{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Chapter 3. 분류"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## 3.1 MNIST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\link\\Anaconda3\\envs\\mlbook\\lib\\site-packages\\sklearn\\utils\\deprecation.py:77: DeprecationWarning: Function fetch_mldata is deprecated; fetch_mldata was deprecated in version 0.20 and will be removed in version 0.22\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n",
      "C:\\Users\\link\\Anaconda3\\envs\\mlbook\\lib\\site-packages\\sklearn\\utils\\deprecation.py:77: DeprecationWarning: Function mldata_filename is deprecated; mldata_filename was deprecated in version 0.20 and will be removed in version 0.22\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'COL_NAMES': ['label', 'data'],\n",
       " 'DESCR': 'mldata.org dataset: mnist-original',\n",
       " 'data': array([[0, 0, 0, ..., 0, 0, 0],\n",
       "        [0, 0, 0, ..., 0, 0, 0],\n",
       "        [0, 0, 0, ..., 0, 0, 0],\n",
       "        ...,\n",
       "        [0, 0, 0, ..., 0, 0, 0],\n",
       "        [0, 0, 0, ..., 0, 0, 0],\n",
       "        [0, 0, 0, ..., 0, 0, 0]], dtype=uint8),\n",
       " 'target': array([0., 0., 0., ..., 9., 9., 9.])}"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.datasets import fetch_mldata\n",
    "\n",
    "mnist = fetch_mldata('MNIST original')\n",
    "mnist"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ 사이킷런에서 읽어 들인 데이터셋들은 일반적으로 비슷한 딕셔너리 구조를 가짐\n",
    "+ 데이터셋을 설명하는 DESCR 키\n",
    "+ 샘플이 하나의 행, 특성이 하나의 열로 구성된 배열을 가진 data 키\n",
    "+ 레이블 배열을 담고 있는 target 키"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'sklearn.utils.Bunch'>\n",
      "dict_keys(['data', 'COL_NAMES', 'target', 'DESCR'])\n",
      "<class 'numpy.ndarray'> <class 'numpy.ndarray'>\n",
      "(70000, 784) (70000,)\n"
     ]
    }
   ],
   "source": [
    "X, y = mnist[\"data\"], mnist[\"target\"]\n",
    "print(type(mnist))\n",
    "print(mnist.keys())\n",
    "print(type(X), type(X))\n",
    "print(X.shape,y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 100 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "array([[7, 3, 5, 2, 2, 1, 1, 8, 4, 1],\n",
       "       [1, 9, 4, 0, 3, 5, 0, 5, 3, 3],\n",
       "       [1, 0, 4, 6, 3, 0, 8, 1, 0, 2],\n",
       "       [0, 5, 5, 8, 3, 2, 3, 5, 9, 6],\n",
       "       [8, 6, 1, 0, 1, 1, 7, 2, 2, 8],\n",
       "       [8, 1, 3, 6, 8, 2, 3, 0, 2, 8],\n",
       "       [3, 4, 5, 9, 3, 6, 1, 0, 7, 7],\n",
       "       [0, 5, 2, 0, 5, 9, 6, 8, 0, 4],\n",
       "       [2, 2, 0, 6, 2, 5, 7, 2, 3, 6],\n",
       "       [6, 8, 7, 7, 8, 2, 3, 0, 5, 2]])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import random\n",
    "\n",
    "target = []\n",
    "plt.figure(figsize=(8,8))\n",
    "\n",
    "for i in range(100):\n",
    "    num = random.randrange(0, 70000)\n",
    "    target.append(int(y[num]))\n",
    "\n",
    "    some_digit = X[num]\n",
    "    some_digit_image = some_digit.reshape(28, 28)\n",
    "    \n",
    "    plt.subplot(10,10,i+1)\n",
    "    plt.imshow(some_digit_image,\n",
    "               cmap = matplotlib.cm.binary,\n",
    "               # cmap = matplotlib.cm.Blues,\n",
    "               interpolation=\"nearest\",)\n",
    "    plt.axis(\"off\")\n",
    "plt.show()\n",
    "\n",
    "target = np.array(target)\n",
    "target = target.reshape(-1, 10)\n",
    "target"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**※ interpolation** 매개변수  \n",
    "https://matplotlib.org/gallery/images_contours_and_fields/interpolation_methods.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "+ MNIST 데이터셋은 이미 훈련 세트(앞쪽 60,000개 이미지)와 테스트 세트(뒤쪽 10,000개 이미지)로 나누어 놓음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = X[:60000], X[60000:], y[:60000], y[60000:]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "+ 훈련 세트를 섞어서 모든 교차 검증 폴드가 비슷해지도록 만듦 (하나의 폴드라도 특정 숫자가 누락되면 안 됨)\n",
    "+ 어떤 학습 알고리즘은 훈련 샘플의 순서에 민감해서 많은 비슷한 샘플이 연이어 나타나면 성능이 나빠짐"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "shuffle_index = np.random.permutation(60000)\n",
    "X_train, y_train = X_train[shuffle_index], y_train[shuffle_index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([4, 7, 8, 0, 5, 3, 9, 6, 1, 2])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.random.permutation(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## 3.2 이진 분류기 훈련"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ **이진 분류기**<sup>binary classifier</sup> (간단한 5-감지기) 타깃 벡터"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_5 = (y_train == 5) # 5는 True고, 다른 숫자는 모두 False\n",
    "y_test_5 = (y_test == 5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ **확률적 경사 하강법**<sup>Stochastic Gradient Descent</sup>(SGD) 분류기\n",
    "+ 매우 큰 데이터셋을 효율적으로 처리 (한 번에 하나씩 훈련 샘플을 독립적으로 처리하기 때문)\n",
    "+ **온라인 학습**에 잘 들어맞음"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SGDClassifier(alpha=0.0001, average=False, class_weight=None,\n",
       "       early_stopping=False, epsilon=0.1, eta0=0.0, fit_intercept=True,\n",
       "       l1_ratio=0.15, learning_rate='optimal', loss='hinge', max_iter=5,\n",
       "       n_iter=None, n_iter_no_change=5, n_jobs=None, penalty='l2',\n",
       "       power_t=0.5, random_state=42, shuffle=True, tol=None,\n",
       "       validation_fraction=0.1, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.linear_model import SGDClassifier\n",
    "\n",
    "sgd_clf = SGDClassifier(max_iter=5, random_state=42)\n",
    "sgd_clf.fit(X_train, y_train_5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5.0 7.0\n"
     ]
    }
   ],
   "source": [
    "some_digit_1, some_digit_2 = X[36000], X[44000]\n",
    "print(y[36000], y[44000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ True, False])"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_clf.predict([some_digit_1, some_digit_2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## 3.3 성능 측정"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.1 교차 검증을 사용한 정확도 측정"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ K-겹 교차 검증"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.9611, 0.9344, 0.9579])"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "cross_val_score(sgd_clf, X_train, y_train_5, cv=3, scoring=\"accuracy\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ 교차 검증 기능을 직접 구현\n",
    "+ 가끔 사이킷런이 제공하는 기능보다 교차 검증 과정을 더 많이 제어해야 할 필요가 있음\n",
    "+ cross_val_score() 함수와 거의 같은 작업을 수행하고 동일한 결과를 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9611\n",
      "0.9344\n",
      "0.9579\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import StratifiedKFold\n",
    "from sklearn.base import clone\n",
    "\n",
    "skfolds = StratifiedKFold(n_splits=3, random_state=42)\n",
    "\n",
    "for train_index, test_index in skfolds.split(X_train, y_train_5):\n",
    "    clone_clf = clone(sgd_clf)\n",
    "    X_train_folds = X_train[train_index]\n",
    "    y_train_folds = (y_train_5[train_index])\n",
    "    X_test_fold = X_train[test_index]\n",
    "    y_test_fold = (y_train_5[test_index])\n",
    "    # print(y_test_fold)\n",
    "    # print(train_index)\n",
    "\n",
    "    clone_clf.fit(X_train_folds, y_train_folds)\n",
    "    y_pred = clone_clf.predict(X_test_fold)\n",
    "    n_correct = sum(y_pred == y_test_fold)\n",
    "    print(n_correct / len(y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**※ StratifiedKFold** 클래스  \n",
    "https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.StratifiedKFold.html"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ 모든 이미지를 무조건 5가 아니라고 예측하는 더미 분류기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.9104 , 0.90615, 0.9124 ])"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.base import BaseEstimator\n",
    "\n",
    "class Never5Classifier(BaseEstimator):\n",
    "    def fit(self, X, y=None):\n",
    "        pass\n",
    "    def predict(self, X):\n",
    "        return np.zeros((len(X), 1), dtype=bool)\n",
    "\n",
    "never_5_clf = Never5Classifier()\n",
    "cross_val_score(never_5_clf, X_train, y_train_5, cv=3, scoring=\"accuracy\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[False],\n",
       "       [False],\n",
       "       [False]])"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.zeros((3, 1), dtype=bool)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 정확도를 분류기의 성능 측정 지표로 선호하지 않는 이유를 보여줌\n",
    "+ 특히 **불균형한 데이터셋**을 다룰 때 더욱 선호되지 않음(어떤 클래스가 다른 것보다 월등히 많은 경우)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.2 오차 행렬<sup>confusion matrix</sup>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ e.g. 숫자 5의 이미지를 3으로 잘못 분류한 횟수 → 오차 행렬의 5행 3열"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[False False False ... False False False] (60000,)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_predict\n",
    "\n",
    "y_train_pred = cross_val_predict(sgd_clf, X_train, y_train_5, cv=3)\n",
    "print(y_train_pred, y_train_pred.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[52559,  2020],\n",
       "       [  912,  4509]], dtype=int64)"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "confusion_matrix(y_train_5, y_train_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 행 : **실제 클래스** , 열 : **예측한 클래스**\n",
    "+ 첫 번째 행 : '5 아님'(**음성 클래스**<sup>negative class</sup>)\n",
    "    + 54,104개를 '5 아님'으로 정확하게 분류 → **진짜 음성**<sup>true negative</sup>\n",
    "    + 나머지 475개는 '5'라고 잘못 분류 → **거짓 양성**<sup>false positive</sup>\n",
    "+ 두 번째 행 : '5'이미지(**양성 클래스**<sup>positive class</sup>)\n",
    "    + 1,966개를 '5 아님'으로 잘못 분류 → **거짓 음성**<sup>false negative</sup>\n",
    "    + 나머지 3,455개는 '5'라고 정확하게 분류 → **진짜 양성**<sup>true positive</sup>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[54579,     0],\n",
       "       [    0,  5421]], dtype=int64)"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_perfect_predictions = y_train_5\n",
    "confusion_matrix(y_train_5, y_train_perfect_predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ **정밀도**<sup>precidion</sup> : 양성 예측의 정확도\n",
    "![Equation3-1](./images/Equation3-1.png)\n",
    "**<center>식 3-1 정밀도</center>**\n",
    "+ TP : 진짜 양성의 수 , FP : 거짓 양성의 수"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ **재현율**<sup>recall</sup>=**민감도**<sup>sensitivity</sup>=**진짜 양성 비율**<sup>true positive rate</sup>(TPR) : 분류기가 정확하게 감지한 양성 샘플의 비율\n",
    "![Equation3-2](./images/Equation3-2.png)\n",
    "**<center>식 3-2 재현율</center>**\n",
    "+ FN : 거짓 음성의 수"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "![Figure3-2](./images/Figure3-2.png)\n",
    "**<center>그림 3-2 오차 행렬</center>**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.3 정밀도와 재현율"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.6906111196201562, 0.831765356945213)"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import precision_score, recall_score\n",
    "\n",
    "precision_score(y_train_5, y_train_pred), recall_score(y_train_5, y_train_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 5로 판별된 이미지 중 69%만 정확\n",
    "+ 전체 숫자 5에서 83%만 감지"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ **F<sub>1</sub> score** : 정밀도와 재현율의 **조화 평균**<sup>harmonic mean</sup>\n",
    "+ 정밀도와 재현율을 하나의 숫자로 만들어 두 분류기를 비교할 때 편리함\n",
    "![Equation3-3](./images/Equation3-3.png)\n",
    "**<center>식 3-3 F<sub>1</sub> 점수</center>**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7546443514644351"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import f1_score\n",
    "\n",
    "f1_score(y_train_5, y_train_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 정밀도와 재현율이 비슷한 분류기에서는 F<sub>1</sub> 점수가 높음\n",
    "+ 상황에 따라 정밀도가 중요할 수도 있고 재현율이 중요할 수도 있음\n",
    "    + 안전한 동영상 걸러내기 → 재현율이 낮더라도 높은 정밀도가 필요\n",
    "    + 감시 카메라 → 정밀도가 낮더라도 높은 재현율이 필요\n",
    "\n",
    "▣ 정밀도를 올리면 재현율이 줄고 그 반대도 마찬가지 → **정밀도/재현율 트레이드오프**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.4 정밀도/재현율 트레이드오프"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ SGDClassifier는 **결정 함수**<sup>decision function</sup>를 사용하여 각 샘플의 점수를 계산함\n",
    "+ 이 점수가 임곗값보다 크면 샘플을 양성 클래스에 할당하고 그렇지 않으면 음성 클래스에 할당함\n",
    "\n",
    "![Figure3-3](./images/Figure3-3.png)\n",
    "**<center>그림 3-3 결정 임곗값과 정밀도/재현율 트레이드오프</center>**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([14833.02796721])"
      ]
     },
     "execution_count": 186,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "some_digit = X[34887]\n",
    "y_scores = sgd_clf.decision_function([some_digit])\n",
    "y_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ True])"
      ]
     },
     "execution_count": 187,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold = 0\n",
    "y_some_digit_pred = (y_scores > threshold)\n",
    "y_some_digit_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([False])"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "threshold = 20000\n",
    "y_some_digit_pred = (y_scores > threshold)\n",
    "y_some_digit_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([-125178.57694668, -595065.79031589, -251025.86971927, ...,\n",
       "        -161936.93412015, -371781.11244939, -664854.75958195]), (60000,))"
      ]
     },
     "execution_count": 273,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_scores = cross_val_predict(sgd_clf, X_train, y_train_5,\n",
    "                             cv=3, method=\"decision_function\")\n",
    "y_scores, y_scores.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 274,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(y_train_pred == (y_scores > 0)).all()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((59848,), (59848,), (59847,))"
      ]
     },
     "execution_count": 201,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import precision_recall_curve\n",
    "\n",
    "precisions, recalls, thresholds = precision_recall_curve(y_train_5, y_scores)\n",
    "precisions.shape, recalls.shape, thresholds.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "slideshow": {
     "slide_type": "-"
    }
   },
   "source": [
    "+ 가능한 모든 임곗값에 대해 정밀도와 재현율을 계산"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ **결정 임곗값에 대한 정밀도와 재현율**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_precision_recall_vs_threshold(precisions, recalls, thresholds):\n",
    "    plt.plot(thresholds, precisions[:-1], \"b--\", label=\"precision\", linewidth=2)\n",
    "    plt.plot(thresholds, recalls[:-1], \"g-\", label=\"Recall\", linewidth=2)\n",
    "    plt.xlabel(\"Threshold\", fontsize=20)\n",
    "    plt.legend(loc=\"center left\", fontsize=20)\n",
    "    plt.xlim([-700000, 700000])\n",
    "    plt.ylim([0, 1])\n",
    "    \n",
    "plt.figure(figsize=(10, 5))\n",
    "plot_precision_recall_vs_threshold(precisions, recalls, thresholds)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "◈ **재현율에 대한 정밀도(PR 곡선)**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_precision_vs_recall(precisions, recalls):\n",
    "    plt.plot(recalls, precisions, \"b-\", linewidth=2)\n",
    "    plt.xlabel(\"Recall\", fontsize=20)\n",
    "    plt.ylabel(\"Precision\", fontsize=20)\n",
    "    plt.axis([0, 1, 0, 1])\n",
    "\n",
    "plt.figure(figsize=(8, 5))\n",
    "plot_precision_vs_recall(precisions, recalls)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.7897087378640777, 0.7502305847629589)"
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_pred_70000 = (y_scores > 70000)\n",
    "precision_score(y_train_5, y_train_pred_70000), recall_score(y_train_5, y_train_pred_70000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.5 ROC 곡선"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ **수신기 조작 특성**<sup>receiver operating characteristic</sup>(ROC) 곡선\n",
    "+ 이진 분류에서 널리 사용\n",
    "+ **거짓 양성 비율**<sup>false positive rate</sup>(FPR)에 대한 **진짜 양성 비율**<sup>true positive rate</sup>(TPR, 재현율의 다른 이름)의 곡선  \n",
    "※ **FPR**<sup>false positive rate</sup> : 양성으로 잘못 분류된 음성 샘플의 비율  \n",
    "※ **TNR**<sup>true negative rate</sup> : 음성으로 정확하게 분류한 음성 샘플의 비율(=**특이도**<sup>specificity</sup>)  \n",
    "※ **FPR = 1 - TNR**\n",
    "+ 따라서 ROC 곡선은 **민감도**(재현율)에 대한 **1 - 특이도** 그래프"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import roc_curve\n",
    "\n",
    "fpr, tpr, thresholds = roc_curve(y_train_5, y_scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def plot_roc_curve(fpr, tpr, label=None):\n",
    "    plt.plot(fpr, tpr, linewidth=2, label=label)\n",
    "    plt.plot([0, 1], [0, 1], 'k--')\n",
    "    plt.axis([0, 1, 0, 1])\n",
    "    plt.xlabel('FPR', fontsize=20)\n",
    "    plt.ylabel('TPR', fontsize=20)\n",
    "\n",
    "plt.figure(figsize=(8, 5))\n",
    "plot_roc_curve(fpr, tpr)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 검은색 점선은 완전한 랜덤 분류기의 ROC 곡선을 뜻함\n",
    "+ 좋은 분류기는 이 점선으로부터 최대한 멀리 떨어져 있어야 함(왼쪽 위 모서리)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "◈ **곡선 아래의 면적**<sup>area under the curve</sup>(AUC)\n",
    "+ 완벽한 분류기는 ROC의 AUC가 1\n",
    "+ 완전한 랜덤 분류기는 ROC의 AUC가 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9619756139834423"
      ]
     },
     "execution_count": 289,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "roc_auc_score(y_train_5, y_scores)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "+ 일반적으로 양성 클래스가 드물거나 거짓 음성보다 거짓 양성이 더 중요할 때 PR 곡선을 사용하고 그렇지 않으면 ROC 곡선을 사용"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 362,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "forest_clf = RandomForestClassifier(n_estimators=10, random_state=42)\n",
    "y_probas_forest = cross_val_predict(forest_clf, X_train, y_train_5,\n",
    "                                   cv=3, method=\"predict_proba\")\n",
    "y_scores_forest = y_probas_forest[:, 1]  # 양성 클래스에 대한 확률을 점수로 사용\n",
    "fpr_forest, tpr_forest, thresholds_forest = roc_curve(y_train_5, y_scores_forest)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 363,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 2)"
      ]
     },
     "execution_count": 363,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_probas_forest.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8, 5))\n",
    "plt.plot(fpr, tpr, \"b:\", linewidth=2, label=\"SGD\")\n",
    "plot_roc_curve(fpr_forest, tpr_forest, \"Random Forest\")\n",
    "plt.legend(loc=\"lower right\", fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 365,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.9619756139834423, 0.9930032118975847)"
      ]
     },
     "execution_count": 365,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(y_train_5, y_scores), roc_auc_score(y_train_5, y_scores_forest)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 366,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([False, False, False, ..., False, False, False]), (60000,))"
      ]
     },
     "execution_count": 366,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train_pred_forest = cross_val_predict(forest_clf, X_train, y_train_5, cv=3)\n",
    "y_train_pred_forest, y_train_pred_forest.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 367,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.983274647887324, 0.8242021767201624)"
      ]
     },
     "execution_count": 367,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "precision_score(y_train_5, y_train_pred_forest), recall_score(y_train_5, y_train_pred_forest)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}